When one student have tossed her coin for 200 times, the sampling distribution model used is
Normal with mean(sub p-hat)= p=0.5; sigma(sub p-hat) = (root)pq/n=0.0354
Her trial proportion of p-hat=.42 is about 2.26 standard deviations below the expected proportion, which is unusual, but not surprising.
Based to the Normal model, the expected sample sizes is low or lower about 1.2% of the time.